Is it possible to consider that cognitive-behavioral therapy is always scientific? Our response looks at this question dealing with a key element of CBT identity from three different angles: historical, epistemological and clinical; by revisiting the past, the present and finally speculating on the future. Is CBT always founded on proof rather than myths? Is CBT a real therapy for serious illnesses? Is every aspect of CBT efficient? True or false? It is urgent to examine each of these questions. Indeed, CBT could disappear, drowned in the emergence of new approaches. How can its future be envisaged? Firstly, in historical terms, the case of little Albert described by Watson and Rayner in 1920 had a founding role in the comprehension of conditioning mechanisms. However, summarizing the mechanism is fastidious, it is difficult to interpret this experiment, and replication studies give rather contradictory results. Despite this, the future of this child, the original patient, has been the object of much speculation which remains unresolved today. There are other myths surrounding CBT: for example, the “bobo doll” experiment (Bandura, Ross & Ross, 1961) which was very controversial; Beck who explained in an article in 1971 (Beck, 1971), without testing his theory, that dreams revealed a patient's schemas better than any external stimulus; and different “waves” of CBT which are still widely disparaged. On an epistemological level, proof of the efficacity of CBT treatment falls willingly on the psychiatric classifications of DSM-V and CIM-10 type. However, these classifications themselves are called into question today in favor of alternatives such as those suggested by the National Institute of Mental Health with the RDoC (Research Domain Criteria) (Insel, 2014) or also the understanding of mental disorders as a system of symptoms in interaction or in a network, “Complex Causal Network Approach”, rather than as linear dimensions or categories. A generic cognitive-behavioral conceptualization of mental disorders would be possible and compatible with this complex approach (Hofmann, 2014). Furthermore, the assessment of the efficacity of the therapy does not only rely on randomized studies. Five other approaches would be useful to develop to improve knowledge of CBT: experimental single case studies, case studies, qualitative studies, process studies and effectiveness studies. In addition, efficacity is not the only goal and it is important not to forget the effectiveness and efficiency of the treatments used. On a clinical level, CBT was first built on conditioning theories and practice based on exposure, generalization and habituation. Today, therapeutic practices are rich, diversified and sometimes even eclectic, creating a certain blur between those which belong specifically to CBT and those which have been borrowed from other approaches. They all appear to be promising, but are they really? The “dodo bird” effect could explain this phenomenon. Developing research on eclectic models, or better still on integrative concepts, would perhaps allow pruning even an opening up to new developments following the examples of the psychotherapy of schemas (Young et al. 2005) and multimodal psychotherapy (Lazarus 1981). Finally, the future is unpredictable, but three areas of reflection could be suggested: let us not forget to integrate the body into our practices, as changing also implies embodying a new way of being. Let us favor modesty as, unlike the weather forecast, change is not predetermined and depends on the ability of the patient to seize all available opportunities at the right time. Psychotherapy is a form of esthetics, as the greatest therapists were fabulous artists before becoming great scientists!